Intraduodenal infusion of hydrophobic bile salts to bile-fistula rats leads within hours to severe hepatocellular necrosis and cholestasis; simultaneous administration of conjugates of ursodeoxycholate, either intraduodenally or intravenously, reduces or prevents liver injury. To evaluate the short-term protective effects of ursodeoxycholate at the cellular level, we incubated primary monolayer cultures of adult rat hepatocytes or freshly isolated washed human erythrocytes for 1 to 240 min with varying defined concentrations of different bile salts in the presence or absence of ursodeoxycholate. Cytolysis was quantified by measuring the release into the medium of cytosolic lactate dehydrogenase (hepatocytes) or hemoglobin (erythrocytes). In both systems, cytolysis increased sigmoidally with increasing bile salt concentration, and the relative toxicity of different bile salts proceeded in the following order: tauroursodeoxycholate was less toxic than taurocholate, which was less toxic than taurodeoxycholate. Taurochenodeoxycholate was more toxic to erythrocytes than taurodeoxycholate; the two were equally toxic to rat hepatocytes. Unconjugated bile salts were more toxic than their conjugates. The addition of tauroursodeoxycholate to taurochenodeoxycholate or taurodeoxycholate led to time-dependent and concentration-dependent reduction or elimination of the toxicity of the more hydrophobic component. Protection was evident within minutes. With respect to hemolysis, at pH 8.5 glyco was less protective than tauroursodeoxycholate, and free ursodeoxycholate was only minimally protective. We conclude that the hepatocytotoxicity of hydrophobic bile salts at millimolar concentrations is markedly reduced in the presence of tauroursodeoxycholate. Conjugates of ursodeoxycholate also prevented disruption of erythrocytes by bile salts, suggesting that protection does not depend on liver-specific pathways of bile salt uptake, compartmentation, transport or metabolism. We hypothesize that ursodeoxycholate interferes with solubilization of membranes by bile salts; in vivo, ursodeoxycholate may improve cholestatic liver disease by reducing the toxicity of endogenous bile salts at the luminal surface of the canalicular plasma membrane. (HEPATOLOGY 1991;14:920–926).